Reliability-Based Analysis and Design of Wide-Span Structures under Stochastic Wind Loads

The objective of this paper is to present a reliability-based method for structural analysis and design under stochastic wind loads so that nominal wind pressure coefficients ${\mathbf{p}}^{\prime *}$ (associated with the covariance matrix ${\mathbf{C}}_{{\mathbf{p}}^{\prime }}$) can be defined for use in the context of the load and resistance factor design (LRFD) method. The novelty of the formulation presented in this paper is based on the approach used to solve the problem of considering simultaneously the wind load as random variables (the maximum mean wind pressure outside the influence of the structure on the life cycle of the facility) and the wind load as a space and time random process defined by the covariance matrix ${\mathbf{C}}_{{\mathbf{p}}^{\prime }}$ (i.e., a second moment approximation for the non-Gaussian process) and the peak factor $q$. The nominal wind pressure coefficients ${\mathbf{p}}^{\prime }$, associated with the covariance matrix ${\mathbf{C}}_{{\mathbf{p}}^{\prime }}$, are defined so that the three sources of wind randomness can be independently considered: (1) the randomness in the peak factor $q$ (which depends on the mean zero-crossing rate, the skewness, and excess kurtosis of the random process and the duration of the wind storm); (2) the randomness in the maximum storm wind mean pressure in the life cycle of the facility, $w$; and (3) the randomness in the wind-load pressure distribution represented by the covariance matrix ${\mathbf{C}}_{{\mathbf{p}}^{\prime }}$. It is shown that by introducing the nominal wind pressure coefficients ${\mathbf{p}}^{\prime }$ and the other random variables in the failure equation, it is possible to (1) obtain a general and objective criterion for selecting the specific response $r_i$ on which the wind-load distribution should be based; and (2) consider the different failure modes, capacity of the members, and target reliability in the definition of the design load pattern and design load factors. Some important results previously obtained by Kaspersky in the load-response-correlation (LRC) method are used and presented in a more general approach. The proposed method (particularly useful for consideration of the stochastic wind loads on wide-span structures such as tall buildings, long-span roofs, towers, and bridges) is applied to didactic examples and to the design of the roof of the Braga Stadium in Portugal.